Pilot Study: The Effects of Cognitive Behavioral Therapy and Financial Assistance on Crime, Violence, and Employment in Allegheny County

The pilot study will test the enrollment process, the baseline survey, the randomization procedure with just the control group and CBT treatment arm, and coursework component of our interventions. Following the successful completion of the pilot study, we will launch the full-scale experiment.

The interventions for the pilot experiment are as follows:

(1) Cognitive Behavioral Therapy: Individuals assigned to the Cognitive Behavioral Therapy (CBT) treatment will be offered access to a new Recovery Oriented Cognitive Therapy (CT-R) program. CT-R is an extension of CBT specifically designed for individuals who may struggle to benefit from the traditional CBT approach and who may not respond to medication and other psychosocial interventions. While the traditional CBT approach addresses positive symptoms (overt changes in thought and behavior), CT-R employs systematic strategies to overcome entrenched problems with motivation, connection, and negative symptoms (decline in normal functioning and withdrawal from daily life). The CT-R program was developed by the ACDHS in collaboration with the Beck Institute, an organization specializing in the development of curriculum and training resources for CBT. The goal of the program is to help participants increase their motivation, reflect on their thinking and aspirations, practice less harmful responses in dangerous situations, and develop more positive behaviors and identities. Individuals assigned to the CBT treatment are invited to attend voluntary group classes where they will receive CT-R. To encourage attendance at the voluntary classes, we will pay participants $50 to attend each class for up to 10 classes. We will closely track compliance throughout the experiment.

The interventions for the full-scale experiment are as follows:

(1) Cognitive Behavioral Therapy: Individuals assigned to the Cognitive Behavioral Therapy (CBT) treatment will be offered access to a new Recovery Oriented Cognitive Therapy (CT-R) program. CT-R is an extension of CBT specifically designed for individuals who may struggle to benefit from the traditional CBT approach and who may not respond to medication and other psychosocial interventions. While the traditional CBT approach addresses positive symptoms (overt changes in thought and behavior), CT-R employs systematic strategies to overcome entrenched problems with motivation, connection, and negative symptoms (decline in normal functioning and withdrawal from daily life). The CT-R program was developed by the ACDHS in collaboration with the Beck Institute, an organization specializing in the development of curriculum and training resources for CBT. The goal of the program is to help participants increase their motivation, reflect on their thinking and aspirations, practice less harmful responses in dangerous situations, and develop more positive behaviors and identities. Individuals assigned to the CBT treatment are invited to attend voluntary group classes where they will receive CT-R. To encourage attendance at the voluntary classes, we will pay participants $50 to attend each class for up to 10 classes. We will closely track compliance throughout the experiment.

(2) Financial Assistance: Individuals assigned to the financial assistance treatment group will be offered unconditional financial grants of $800 per month for up to 3 months. The goal of this treatment is to help solve participants' immediate financial needs, such as short-term housing and transportation, as they make the difficult transition back into the community after jail.

2025-12-01

2026-04-01

The anticipated primary outcomes in the full experiment will include three administrative outcomes:

(1) Recidivism within two years of enrolling in the experiment, which we measure using an indicator equal to 1 if the individual is rebooked into the Allegheny County Jail within two years of enrolling in the experiment.

(2) Violence involvement within two years of enrolling in the experiment, which we measure using an indicator equal to 1 if the individual receives a violence charge (e.g., aggravated assault, homicide, or murder of any degree) or goes to the ER with a gunshot wound within two years of enrolling in the experiment,

(3) Non-employment within two years of enrolling in the experiment, which we measure using the number of quarters during which the individual is not employed in the two years after enrolling in the experiment.

The anticipated secondary outcomes in the full experiment will include:

(1) The three primary administrative outcomes (recidivism, violence involvement, and non-employment) measured within two years of enrolling in the experiment, for subgroups delineated by ex-ante recidivism risk predictions based on machine learning algorithms and human assessments.

(2) The three primary administrative outcomes measured within five years of enrolling in the experiment using surrogate indices based on adverse administrative outcomes measured within two years of enrollment in the experiment.

(3) The adverse administrative outcomes used to construct the surrogate indices and measured within two years of enrolling in the experiment. These outcomes include homeless housing utilization, wages, public benefits receipt, child welfare involvement, inpatient hospitalization, emergency department utilization, and behavioral health treatment utilization.

In the full-scale experiment, we will conduct a randomized controlled trial in Allegheny County to test two conceptually different ways to assist individuals exiting jail and transitioning back into the community. We will study the effectiveness of our interventions at reducing recidivism, violence involvement, and non-employment among individuals recently released from jail. We are starting with a pilot study to test the enrollment process, the baseline survey, the randomization procedure with just the control group and CBT treatment arm, and CBT coursework. Following the completion of the study, we will launch the full-scale experiment and provide additional details on the experimental design.

Not available

Randomization will be done in an office by a computer for both the pilot study and full-scale experiment.

The unit of randomization is the individual for both the pilot study and full-scale experiment. Eligible individuals exiting the Allegheny County Jail will be randomized into a control group or the CBT treatment arm in the pilot, and into a control group or one of two treatment groups in the full experiment.

No

Around 3,100 individuals in the full experiment

Around 3,100 individuals in the full experiment

The probability of being assigned to each treatment arm will be finalized after the pilot and before the full experiment. Assuming reasonable enrollment during the pilot, we will allocate 1,300 individuals to the control group, 1,300 individuals to the CBT treatment group, and 500 individuals to the financial assistance treatment group.

Our power analysis assumes a sample size of 3,100 participants, 50% take-up in the CBT treatment group, and 100% take-up for the financial assistance treatment group. We allocate 1,300 participants to the control group, 1,300 participants to the CBT treatment group and 500 to the financial assistance treatment group. We compute the minimum detectable effects (MDE) for an intent-to-treat analysis. We assume power of 80% with a significance level of 5%. We find MDEs on jail re-entry of 11.0 percentage points (21%) for the CBT treatment group and 7.4 percentage points (14%) for the financial assistance treatment group. We find MDEs on violence involvement of 8.6 percentage points (45%) for the CBT treatment group and 5.8 percentage points (30%) for the financial assistance treatment group. We find MDEs on non-employment of 0.57 quarters (28%) for the CBT treatment group and 0.38 quarters (19%) for the financial assistance treatment group. The MDEs do not account for the improved precision that can be achieved by controlling for the pre-treatment covariates that we will include in our final analysis.